The present invention relates to an apparatus for the removal of ghost in television signals used in television receivers and video receivers that handle television signals.
FIG. 1 shows one example of a conventional ghost removal apparatus. The processing in this embodiment is all performed digitally.
The numeral 1 represents a transversal filter that has the television signals X(t) as the input, and which compensates for waveform distortion and removes the ghost component. In this embodiment, the filter is configured from a digital filter that has variable filter coefficients. Such a transversal filter 1 is set with a batch of filter coefficients that is of the opposite phase to a ghost component that is included in the input signals, for example. The output of the transversal filter 1 becomes the output image signals Y (t) and is also supplied to a waveform extraction circuit 2. The waveform extraction circuit 2 extracts reference signals inclined in predetermined periods in the television signals. These reference signals are of a pattern from which a reference timing in a horizontal scan can be detected, and FIG. 2 shows typical examples of a reference signal.
FIG. 2(a) and FIG. 2(b) respectively represent where step-shaped signals are partially overlapped, with FIG. 2 (a) showing where the rising edge for example, of a step-shaped signal is used as the reference timing and FIG. 2 (b) showing where the falling edge for example, of a step-shaped signal is used as the reference timing.
FIG. 2(c) is a pulse signal that is overlapped on a vertical synchro signal, and either the rising edge or the falling edge is used as the reference in this case as well.
In the 8-field sequence method, these reference signals (data) described above are obtained by subtraction from the data of a predetermined other field, of the data from a predetermined other field.
The above described specific pattern signals extracted by the waveform extraction circuit 2 are supplied to an adding & averaging circuit 3 and the adding & averaging circuit 3 smooths the noise components that are generated randomly in the output of the waveform extraction circuit 2. FIG. 3 shows a specific example of the circuits for the waveform extraction circuit 2 and the adding & averaging circuit 3.
In FIG. 3, the waveform extraction circuit 3 has a gate circuit 21 that passes an input signal which is applied to it at predetermined periods that include the reference signals, and the gate circuit 21 has its on and off controlled by control signals from a controller that is not indicated in the figure. The signals having a predetermined period and that have passed the gate circuit 21 are supplied to a memory 32 via an adder 31 of the adding & averaging circuit 3 and furthermore, the output of the memory 32 is fed back to the adder 31. More specifically, the adder 31 adds the signals that have been given a predetermined time delay and which have already been written in the memory 32, and supplies them to the memory 32 so that noise components that exist randomly in the direction of the time axis, for the output of the memory 32 is smoothed.
After this adding & averaging processing has been performed for a plural number of times, the gate circuit 33 becomes open in accordance with a take-in pulse supplied from the controller and the signals that have been added and smoothed are taken in.
Returning again to FIG. 1, the output of the adding & averaging circuit 3 has the influence of the DC portion removed by a differentiating circuit 4 and then, is compared with a reference waveform signal that is generated in a reference waveform generation circuit 5, with a subtractor 6 being used as the means of comparison. The output of the subtractor 6 is supplied to a magnification setting circuit 7 as error signals. The magnification setting circuit 7 determines the loop magnification and this set magnification is used as the basis for a coefficient setting circuit 8 to generate coefficient data in order to set the batch of filter coefficients of the transversal filter 1 described above.
In the manner described above, the configuration of a ghost removal circuit formed by a closed loop comprising the transversal filter 1, the waveform extraction circuit 2, the adding & averaging circuit 3, the differentiating circuit 4, the subtractor 6, the reference waveform generation circuit 5, the magnification setting circuit 7 and the coefficients setting circuit 8. More specifically, the filter coefficients of the transversal filter 1 change successively in accordance with the error signals that are obtained from the reference signals derived from the input television signals and the reference waveform signals generated in the circuit 5, and control of the closed loop is performed so as to remove the ghost component.
The following is a description of another example of a conventional ghost removal apparatus, with reference to FIG. 4.
The ghost removal apparatus according to the other example is the same as the apparatus described with reference to FIG. 1, except that the adding & averaging circuit 3 and the differentiating circuit 4 are absent, and hence duplication of description will be avoided by using corresponding numerals to represent corresponding portions.
The waveform extraction circuit 2 extracts the predetermined periods that include the specific pattern signals within the television signals, and in accordance with necessity, implements a predetermined waveform conversion such as differentiation, and outputs the signals. This specific pattern signal is a pattern that is used to detect and obtain the reference timing in a horizontal scan period, and FIG. 5 shows typical examples of this specific pattern signal. FIG. 5(a) and FIG. 5 (b) show where respective step-shaped signals overlap one portion of a horizontal scan period, with FIG. 5(a) showing the case where the rising edge of a step-shaped signal is used as the reference timing and FIG. 5(b) showing the case where the falling edge of a step-shaped signal is used as the reference timing. For example, when the falling edge portion of the step-shaped signal of FIG. 5(b) is differentiated, a pulse-shaped signal such as that shown in FIG. 5(c) is obtained. FIG. 5(dis a step-shaped signal that has a narrow width and which overlaps the vertical synchro signal, and in this case as well, it is possible to use either the rising edge or the falling edge as the reference timing.
When a specific pattern signal extracted by the waveform extraction circuit 2 has waveform conversion processing such as differentiation or the like implemented, the subtractor 6 is used as the means of comparison to compare the reference signal that is generated from the reference waveform generation circuit 5. The output of the subtractor 6 is supplied to the magnification setting circuit 7 as the error signal. The magnification setting circuit 7 determines the magnification in order to determine the filter coefficients that are set in the transversal filter 1, and the magnification set in the magnification setting circuit 7 is used as the basis for the coefficient setting circuit 8 to generate the coefficient data that is used to set the filter coefficients of the transversal filter 1.
In the manner as has been described above, the ghost removal circuit is formed by a closed loop comprising the transversal filter 1, the waveform extraction circuit 2, the subtractor 6, the reference waveform generation circuit 5, the magnification setting circuit 7 and the coefficient setting circuit 8. More specifically, it is a circuit that obtains the reference timing from the specific pattern signals in the input television signals, performs successive changing of the filter coefficients of the transversal filter 1 in accordance with error signals obtained from the reference timing and the reference signals, and performs closed loop control so that the ghost components are removed.
Both of the two examples of the conventional technology described above are configured by a feedback loop and are ghost removal circuits comprising a feed-forward loop configured by disposing a waveform extraction circuit 2 on the input side of the transversal filter 1.
However, when there is a favorable S/N (signal to noise ratio) for the input television signals in the case of the former of the conventional apparatus and as shown in FIG. 6 (a), there is a certain degree of noise removal performed by the adding & averaging processing but as shown in FIG. 6 (b), in the status where the input television signals have a poor S/N, it is not possible to sufficiently suppress the noise by using only the adding & averaging circuit 3, and so it becomes impossible to correctly detect the ghost components that are included in the reference signals.
In addition, when pulse-type noise enters during the predetermined period that includes the reference signal, and when the waveform extraction circuit 2 extracts signals other than those of the predetermined periods and adding processing is performed in that state, a component that is not a ghost component is regarded as a ghost.
In either case, the error signals become the output from the subtractor 6 and ghost removal processing corresponding to this error data is performed to therefore cause problems with the output image.
Also, the coefficient data that are set by the coefficient setting circuit 8 in the case of the latter example of conventional technology has no band limit and so the frequency band component that has the coefficient data also has a component (high region component) other than the frequency band component that has the reference signals and there are cases where this component which is outside of the band causes the transversal filter 1 to perform erroneous filter action. In cases such as these, there is the disadvantage that the convergence to the ghost removal status is slowed.